Envoyé par
Metalink
Any change in the hostname of a machine, requires modifications to the
SQL*Net configuration files and a restart of the Intelligent Agent.
Secondly, all machines connected with the agent, must be able to resolve this
new hostname to the same IP address. This requires additional changes made on
the machines running the OEM V1.X console, or the Oracle Management
Server for OEM V2.X.
The changes include a remove of the agent 'Q' files. These file contain the
details of the jobs and events registered towards the services on this
machine.
Steps
-----
When the IP address of a machine changes, there a few things to do:
1) Start the Enteprise Manager console
* From the Job pane, cancel all pending jobs for that destination.
- Select the 'Active' job tab
- Select a job which is registered towards the node and delete
* From the Event pane, cancel all registered events for that destination.
- Select the 'Registrations' event tab
- De-register all events registered towards the node
After these modifications, the agent is now no longer used for monitoring.
Cleanup of the agent files can now happen.
2) On the server, clean up the intelligent agent files.
* Stop the Intelligent Agent.
On UNIX: lsnrctl dbsnmp_stop
On NT : Stop the agent via the Services applet of the control panel
Start->Settings->Control Panel->Services->OracleAgent
* From the $ORACLE_HOME/network/agent directory, remove all files with the
'Q' extention.
On UNIX: $ rm *.q
On NT : C:\ > del *.q
* Verify the SQL*Net configuration.
Since the hostname has changed, the SQL*Net configuration needs to be
adjusted to make sure the connections are now accepted for the new
hostname.
3) Drop and rediscover the node
If only the IP address of the machine has changed, but the hostname has
remained the same, you do not have to drop the node from the console.
If the name of the machine has changed, the services on that node have
to be rediscovered in order to monitor them again.
- From the navigator tree, expand the nodes, and delete the node from the
treelist
- Discover the node again via the 'Service Discovery' menu
- After the discovery, verify the preferred credentials for all services
on that node.
4) Then reschedule the jobs, and re-register the events
After the agent is online again, and all services have been rediscovered,
the jobs and events can be registered again on that node.
For examples, see:
<Note:73989.1> Steps to Create Job in Oracle Enterprise Manager
<Note:2064006.4> Steps for Creating Events and Fixit Jobs
Explanation
-----------
The agent uses the IP address of the machine it is running on as an encryption
key for the 'Q' files. This means that the 'Q' files are unique for that IP
address and will only be used for that IP address.
On startup, if the 'Q' files exist, the agent verifies the key found in the
files with the IP address it has detected during the discovery. If this
information does not match, these errors will be generated:
NMS-302, Failed to initialize user queue
NMS-351, Encryption key supplied is not the one used to encrypt file
For job and event information, the specific IP address is known and 'listened'
for by the OEM V1.X daemon manager, or the Oracle Management server in OEM V2.X.
If an agent is started on a different IP address, the awaited information will
never be transmitted back because the IP address is not known. This results
in the status of those jobs and/or events will not being updated at the OEM
console.
Thank you for using our forum
Nahed Majzoub
Partager